Opinions expressed by Entrepreneur contributors are their own.
Key Takeaways
- Sophistication is assumed to protect people from being fooled. Our data found the opposite. Higher income and more education both correlated with worse accuracy at distinguishing an AI-edited video from a human-edited video.
- People weren’t guessing randomly. They used a consistent, wrong heuristic: Simple editing must mean a human made it; polished editing must mean AI made it. In this study, that was backwards.
- This goes beyond video. Any business assuming a certain customer segment will catch a shortcut or a quality drop is relying on an intuition that has probably never been tested.
- The fix isn’t guessing which audience is more discerning. It’s testing the actual content or product against real people and watching what they do.
Most businesses assume their most educated, highest-earning customers are also their most discerning ones — the group least likely to be fooled by a shortcut, a shallow claim, or a piece of content that only looks authentic.
We ran a study that put that assumption to the test, and the data went the other way. It is a small reversal on paper, but it points to a bigger mistake a lot of businesses make without realizing it.
The assumption almost every business makes
Ask most founders which of their customers would catch a fake, a cut corner or an AI-generated shortcut, and they will usually point to the same group. The educated ones. The higher earners. The audience segment assumed to be paying closer attention and harder to fool.
This assumption quietly shapes a lot of real decisions. Which segment gets the polished version of something and which gets the rushed one. Which audience a company worries about disappointing and which one it assumes will not notice the difference. Nobody usually says this out loud, but it sits underneath a lot of choices about where to invest extra care and where to cut corners.
We had never actually tested whether that assumption holds up. Most businesses never do. It is treated as common sense, the kind of thing nobody questions because it feels obviously true. So we built a study to find out if it actually was.
What actually happened when we tested it
We showed 315 people the same raw footage edited two different ways, once by a professional human editor, once entirely by AI, and asked them to guess which was which. The full methodology and dataset are published here. When we broke the results down by income and education, the pattern ran opposite to what most people would predict.
Accuracy went down as household income went up. The highest earners in the sample were the least likely to correctly identify the human-edited video. Education showed a similar pattern. Respondents with a graduate degree were less accurate, at 34.8% correct, than those with a bachelor’s degree at 51%, or even a high school education at 45.5%.
The group most businesses assume would be hardest to fool turned out to be the easiest. Not by a small margin either. The gap between the most and least accurate education groups was wide enough that it would change how a business should think about which audience to be most careful with.
Why the smartest guesses were the most wrong
The reasoning behind the wrong guesses was remarkably consistent across the sample. Most people assumed simpler editing meant a human made it, and more polish and more graphics meant AI made it. In this case, that assumption ran exactly backwards. Our human editor was the one who added the branded motion graphics and captions. The AI kept things minimal.
What stood out is that more educated and higher-earning respondents did not apply a different rule than everyone else. They applied the same wrong rule, just with more confidence. Sophistication did not lead people to question the shortcut. It led them to trust their first read more, and their first read happened to be wrong.
This is a useful reminder for anyone running a business on assumptions about which part of their audience is paying closest attention. Confidence and analytical ability are not the same thing as accuracy. Sometimes a sharper audience just applies a wrong assumption more consistently than a less sharp one would, which makes the mistake look more deliberate than it actually is. It is not that sophistication stopped mattering. It is that it got pointed at the wrong signal, just as confidently as everything else these respondents were probably right about most days.
What this means for how you test assumptions
The specific finding here is about video editing, but the underlying lesson is not. Any business that assumes a certain customer segment is too smart to be fooled, too experienced to miss a shortcut or too sophisticated to fall for surface-level polish is relying on an intuition that has probably never been tested against real behavior.
The fix is not guessing harder about which audience is more discerning. It is testing the actual thing, the actual content, the actual product, against real people and watching what they actually do, not who you assumed would catch it. We ran the full study and published the complete dataset, including the parts that surprised us, because the honest results were more useful to us than the comfortable assumption would have been.
Most businesses only find out an assumption like this was wrong after it costs them something: a lost client, a damaged reputation, a decision made on a hunch that never held up. Testing it first is cheaper every time, and it usually takes less effort than defending the wrong assumption after the fact.
Key Takeaways
- Sophistication is assumed to protect people from being fooled. Our data found the opposite. Higher income and more education both correlated with worse accuracy at distinguishing an AI-edited video from a human-edited video.
- People weren’t guessing randomly. They used a consistent, wrong heuristic: Simple editing must mean a human made it; polished editing must mean AI made it. In this study, that was backwards.
- This goes beyond video. Any business assuming a certain customer segment will catch a shortcut or a quality drop is relying on an intuition that has probably never been tested.
- The fix isn’t guessing which audience is more discerning. It’s testing the actual content or product against real people and watching what they do.
Most businesses assume their most educated, highest-earning customers are also their most discerning ones — the group least likely to be fooled by a shortcut, a shallow claim, or a piece of content that only looks authentic.
We ran a study that put that assumption to the test, and the data went the other way. It is a small reversal on paper, but it points to a bigger mistake a lot of businesses make without realizing it.
The assumption almost every business makes
Ask most founders which of their customers would catch a fake, a cut corner or an AI-generated shortcut, and they will usually point to the same group. The educated ones. The higher earners. The audience segment assumed to be paying closer attention and harder to fool.